44
Shallow water equations in channel networks Maya Briani Istituto per le Applicazioni del Calcolo \M. Picone" Consiglio Nazionale delle Ricerche [email protected] http://www.iac.rm.cnr.it/ briani May 5 - 2021 Maya Briani (IAC-CNR) Shallow water equations in channel networks May 5 - 2021 1 / 42

Shallow water equations in channel networks

  • Upload
    others

  • View
    4

  • Download
    0

Embed Size (px)

Citation preview

Shallow water equations in channel networks

Maya Briani

Istituto per le Applicazioni del Calcolo “M. Picone”Consiglio Nazionale delle Ricerche

[email protected]

http://www.iac.rm.cnr.it/∼briani

May 5 - 2021

Maya Briani (IAC-CNR) Shallow water equations in channel networks May 5 - 2021 1 / 42

Plan of the talk

(1) The Shallow Water Equations

(2) The Riemann Problem

(3) Water Flow in Canal Network

(4) The Junction Riemann Problem

(5) Fluvial to torrential transition

(6) Conclusions and Open Problems

Maya Briani (IAC-CNR) Shallow water equations in channel networks May 5 - 2021 2 / 42

The Shallow Water Equations

The Shallow Water Equations

The one-dimensional shallow water equations describe the waterpropagation in a canal with rectangular cross-section and constant slope:

∂th+ ∂x(hv) = 0 conservation of mass

∂t(hv) + ∂x(hv2 + 1

2gh2) = 0 conservation of momentum

(1)

I h(x, t) the water height

I v(x, t) the water velocity at time t and location x along the canal

I g the gravity constant

For the purpose of this talk, we have assumed a steady state friction on allcanals and horizontal canals with zero slope.

Maya Briani (IAC-CNR) Shallow water equations in channel networks May 5 - 2021 3 / 42

The Shallow Water Equations

The Shallow Water Equations

We reformulate system (1) in vector form as

∂tu+ ∂xf(u) = 0 (2)

where

u =

(hq

)f(u) =

(hv

hv2 + 12gh

2

)(3)

and q = hv (discharge, it measures the flow rate of water past a point).

The Riemann Problem∂tu+ ∂xf(u) = 0,

u(x, 0) =

{ul if x < 0,ur if x > 0.

(4)

Here u(x, 0) = (h(x, 0), q(x, 0)) and ul = (hl, ql) and ur = (hr, qr).

Maya Briani (IAC-CNR) Shallow water equations in channel networks May 5 - 2021 4 / 42

The Riemann Problem for shallow water equations

The Riemann Problem

x

t

0

l-wave r-waveu∗

ul ur↗ ↖

Figure: The solution to the Riemann problem. The intermediate state u∗ is constant in theregion delimited by l-wave and r-wave. l- and r-waves are shocks or rarefactions.

I The solution to this Riemann problem consists of the l-wave and the r-waveseparated by an intermediate state u∗ = (h∗, q∗).

I This intermediate state is connected to ul = (hl, ql) through a physicallycorrect l-waves, and to ur = (hr, qr) through a physically correct r-wave.

Maya Briani (IAC-CNR) Shallow water equations in channel networks May 5 - 2021 5 / 42

The Riemann Problem for shallow water equations

The Shallow Water Equations - The Riemann Problem

For smooth solution, system (2) can equivalently be written in thequasilinear form

∂tu+A(u)∂xu = 0where the Jacobian matrix A(u) = f ′(u) is

A(u) =

(0 1

−v2 + gh 2v

)The eigenvalues of the matrix A(u) are

λ1(u) = v −√gh, λ2(u) = v +

√gh

with the corresponding eigenvectors r1(u) = (1, v +√gh)T and

r2(u) = (1, v +√gh).

Maya Briani (IAC-CNR) Shallow water equations in channel networks May 5 - 2021 6 / 42

The Riemann Problem for shallow water equations

The Riemann Problem

I The shallow water equations are genuinely nonlinear (∇λj(u) · rj(u) 6= 0,j = 1, 2) and so the Riemann problem always consists of two waves, each ofwhich is a shock or rarefaction.

I The left and right characteristics are associated to λ1 and λ2 respectively.

I λ1 = v −√gh and λ2 = v +

√gh can be of either sign.

I The ratio Fr = |v|/√gh is called the Froude number.

I When v = q/h is smaller than the speed√gh of the gravity waves:

|v| <√gh or Fr < 1

the fluid is said to be fluvial or subcritical.

If |v| >√gh the fluid is said to be torrential or supercritical.

I Under the fluvial regimeλ1 < 0 λ2 > 0

and there will be one left (with negative speed) and one right (with positivespeed) going wave.

Maya Briani (IAC-CNR) Shallow water equations in channel networks May 5 - 2021 7 / 42

The Riemann Problem for shallow water equations

The solution always consists of two waves, each of which is a shock orrarefaction:

(R) Centered Rarefaction Waves. Assume u+ lies on the positivei-rarefaction curve through u−, then we get

u(x, t) =

u− for x < λi(u

−)t,Ri(x/t;u

−) for λi(u−)t ≤ x ≤ λi(u+)t,

u+ for x > λi(u+)t,

(S) Shocks. Assume that the state u+ is connected to the right of u− byan i-shock, then calling λ = λi(u

+, u−) the Rankine-Hugoniot speedof the shock, the function

u(x, t) =

{u− if x < λtu+ if x > λt

provides a the solution to the Riemann problem. For strictlyhyperbolic systems, we have that

λi(u+) < λ(u−, u+) < λi(u

−), λ(u−, u+) =q+ − q−

h+ − h−.

Maya Briani (IAC-CNR) Shallow water equations in channel networks May 5 - 2021 8 / 42

The Riemann Problem for shallow water equations

The Riemann Problem - Lax curves

To find the intermediate state u∗ in general we can define two functions φland φr by

φl(h) =

{vl − 2(

√gh−

√ghl) if h < hl (rarefaction)

vl − (h− hl)√g h+hl2hhl

if h > hl (shock wave),

and

φr(h) =

{vr + 2(

√gh−

√ghr) if h < hr (rarefaction)

vr + (h− hr)√g h+hr2hhr

if h > hr (shock wave).

For a given state h

I the function φl(h) returns the value of v such that (h, hv) can be connectedto ul by a physically correct l-wave

I the function φr(h) returns the value of v such that (h, hv) can be connectedto ur by a physically correct r-wave.

I So, h∗ is such that φl(h∗) = φr(h

∗)

Maya Briani (IAC-CNR) Shallow water equations in channel networks May 5 - 2021 9 / 42

The Riemann Problem for shallow water equations

The Riemann Problem

h

0 1 2 3 4 5 6 7

v

-10

-5

0

5

10

15

λ1 < 0 λ2 > 0

uluru∗

φl

Rl

Sl

φrSr

Rr

Maya Briani (IAC-CNR) Shallow water equations in channel networks May 5 - 2021 10 / 42

The Riemann Problem for shallow water equations

Example: Dam-Break and Riemann Problem

Consider the Riemann problem with

ul =

(hlql

)=

(10

)ur =

(hrqr

)=

(0.50

).

I hl > hr and ql = qr = 0. This Riemann problem models what happens in adam separating two levels of water breaks at time t = 0

I The solution consists of a l-rarefaction and a r-shock

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1

x

0

0.2

0.4

0.6

0.8

1

h

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1

x

0

0.2

0.4

0.6

0.8

1

q

Maya Briani (IAC-CNR) Shallow water equations in channel networks May 5 - 2021 11 / 42

Water Flow in a Channel Network

Water Flow in a Canal Network

Maya Briani (IAC-CNR) Shallow water equations in channel networks May 5 - 2021 12 / 42

Water Flow in a Channel Network

Water Flow in a Channel Network

−→

JCanal 1

Canal 2

Canal 3On each canal

∂tui + ∂xf(ui) = 0

?y#

Maya Briani (IAC-CNR) Shallow water equations in channel networks May 5 - 2021 13 / 42

Water Flow in a Channel Network

Water Flow in a Channel Network

−→

JCanal 1

Canal 2

Canal 3On each canal

∂tui + ∂xf(ui) = 0

?y#

Maya Briani (IAC-CNR) Shallow water equations in channel networks May 5 - 2021 13 / 42

Water Flow in a Channel Network

Water Flow in a Channel Network

−→

JCanal 1

Canal 2

Canal 3On each canal

∂tui + ∂xf(ui) = 0

?y#

Maya Briani (IAC-CNR) Shallow water equations in channel networks May 5 - 2021 13 / 42

Water Flow in a Channel Network

Water Flow in a Channel Network

−→

JCanal 1

Canal 2

Canal 3On each canal

∂tui + ∂xf(ui) = 0

unknowns:

u∗1 = (h∗1, q

∗1)

u∗2 = (h∗2, q∗2)

u∗3 = (h∗3, q∗3)y

#

Maya Briani (IAC-CNR) Shallow water equations in channel networks May 5 - 2021 14 / 42

Water Flow in a Channel Network

Water Flow in a Channel Network

−→

JCanal 1

Canal 2

Canal 3On each canal

∂tui + ∂xf(ui) = 0

Conservation of mass

Junction Riemann Problem

Closure conditionsy#

Maya Briani (IAC-CNR) Shallow water equations in channel networks May 5 - 2021 15 / 42

Water Flow in a Channel Network

Water Flow in Canal Network: 1-to-2 Junction

Assuming that the three canals are connected at x = 0:

Canal 1 (x < 0)

∂tu1 + ∂xf(u1) = 0

Canal 2 and 3 (x > 0)

∂tu2 + ∂xf(u2) = 0

∂tu3 + ∂xf(u3) = 0

Assuming the conservation of mass

q∗1(0−, t) = q∗2(0

+, t) + q∗3(0+, t)

To get a well-posed problem we need 5 additional conditions

Maya Briani (IAC-CNR) Shallow water equations in channel networks May 5 - 2021 16 / 42

Water Flow in a Channel Network

The Junction Riemann Problem

The solution is determined once one assigns a Riemann Solver at thejunction. Considering only subcritical states

I given constant initial conditions (u0i , u0j ) (i ranges over incoming

canals, j over outgoing ones);

I the Junction Riemann solution consists of intermediate states (u∗i , u∗j )

satisfying some other junction conditions

x

t

0

l-wave r-wave

u0i u0j

u∗i u∗j↗ ↖←→

junction conditions

Maya Briani (IAC-CNR) Shallow water equations in channel networks May 5 - 2021 17 / 42

Water Flow in a Channel Network

Left-half Riemann Problem (the case of an incoming canal)

We fix a left state and we look for the right states attainable by waves ofnegative speed. ⇒ Fix ul = (hl, ql), we look for the set of pointsu∗l = (h∗l , q

∗l ) such that the solution to the Riemann problem

∂tu+ ∂xf(u) = 0,

u(x, 0) =

{ul if x < 0u∗l if x > 0

contains only waves with negative speed (λ1(u∗l ) < 0).

x

t

0

l-wave

ul

u∗l↗

0 1 2 3 4 5 6 7

h

-10

-5

0

5

10

15

v

φl

ul

λ1 < 0

Maya Briani (IAC-CNR) Shallow water equations in channel networks May 5 - 2021 18 / 42

Water Flow in a Channel Network

Right-half Riemann problem (the case of an outgoingcanal)

We fix a right state and we look for the left states attainable by waves ofpositive speed.⇒ Fix ur = (hr, qr), we look for the set of pointsu∗r = (h∗r , q

∗r ) such that the solution to the Riemann problem

∂tu+ ∂xf(u) = 0,

u(x, 0) =

{u∗r if x < 0ur if x > 0

contains only waves with positive speed (λ2(u∗r)) > 0).

x

t

0

r-wave

ur

ur↖

0 1 2 3 4 5 6 7

h

-10

-5

0

5

10

15

v

φr

ur

λ2 > 0

Maya Briani (IAC-CNR) Shallow water equations in channel networks May 5 - 2021 19 / 42

Water Flow in a Channel Network

The Junction Riemann Problem

0 1 2 3 4 5 6 7

h

-10

-5

0

5

10

15

v

φl

ul

λ1 < 0

0 1 2 3 4 5 6 7

h

-10

-5

0

5

10

15

v

φr

ur

λ2 > 0

v∗l = φl(h∗l ;hl, vl) v∗r = φr(h

∗r ;hr, vr)

|v∗l,r| <√gh∗l,r

Maya Briani (IAC-CNR) Shallow water equations in channel networks May 5 - 2021 20 / 42

Water Flow in a Channel Network

Water Flow in a Canal Network - Junction Conditions

We have so far set 4 conditions:

q∗1 = q∗2 + q∗3 and

v∗1 = φl(h∗;u01) v∗2 = φr(h

∗;u02) v∗3 = φr(h∗;u03)

We need 2 additional conditions:

I Physical reasons motivate different choices of conditions that areoriginally derived by engineers

I Which conditions are used often depends on if the flow is subcriticalor supercritical

Maya Briani (IAC-CNR) Shallow water equations in channel networks May 5 - 2021 21 / 42

Water Flow in a Channel Network

Water Flow in Canal Network - Junction Conditions

The conservation of mass is usually coupled with the following:

I Equal water pressure (equal water heights)

1

2gh2k =

1

2gh2l ∀t > 0

I Energy continuity (equal of energy levels)

hk +v2k2g

= hl +v2l2g

∀t > 0

Other conditions which depend on the geometry:I Preprint 2021 M. Briani, G. Puppo, M. Ribot, Angle dependence incoupling conditions for shallow water equations at canal junctions

Maya Briani (IAC-CNR) Shallow water equations in channel networks May 5 - 2021 22 / 42

Water Flow in a Channel Network

Water Flow in a Channel Network

−→

On each canal∂tui + ∂xf(ui) = 0

h∗1 = h∗

2 = h∗3 = h∗,

v∗1 = v∗2 + v∗3,

v∗1 = φl(h∗;u0

1),

v∗2 = φr(h∗;u0

2),

v∗3 = φr(h∗;u0

3),

|v∗k| <√gh∗

#

starting by Riemann data

u1(J−) = u01, u2(J

+) = u02, u3(J+) = u03.

Maya Briani (IAC-CNR) Shallow water equations in channel networks May 5 - 2021 23 / 42

Water Flow in a Channel Network

The Junction Riemann Problem

The solution at the Junction then consists on solving the non-linearsystem

h∗1 = h∗2 = h∗3 = h∗,

v∗1 = v∗2 + v∗3,

v∗1 = φl(h∗;u01),

v∗2 = φr(h∗;u02),

v∗3 = φr(h∗;u03),

(5)

with h∗ > 0 and the subcritical assumption |v∗k| <√gh∗, k = 1, 2, 3

I the system admits a unique solution (see for instance Marigo 2010)... but the solution not always verifies the subcritical condition

I suitable initial data have to be given to ensure the fluvial regime tothe problem.

Maya Briani (IAC-CNR) Shallow water equations in channel networks May 5 - 2021 24 / 42

Water Flow in a Channel Network

Fluvial and Torrential regime

What happens if we expand the domain to include the torrential regime?

0 5 10 15

h

-25

-20

-15

-10

-5

0

5

10

15

20

25

v

λ1 < 0 λ2 > 0

λ1 > 0 λ2 > 0

λ1 < 0 λ2 < 0

v =√gh

v = −√gh

0 5 10 15

h

-200

-150

-100

-50

0

50

100

150

200

q

λ1 < 0 λ2 > 0

λ1 > 0 λ2 > 0

λ1 < 0 λ2 < 0

What happens if one of the states is in the torrential regime?

Maya Briani (IAC-CNR) Shallow water equations in channel networks May 5 - 2021 25 / 42

Water Flow in a Channel Network

Fluvial and Torrential regime: the case of an incomingcanal

λ1 < 0

λ1 > 0

λ1 < 0

q = h√gh

hφl(h;u0i )

admissible junction values

u0i

Maya Briani (IAC-CNR) Shallow water equations in channel networks May 5 - 2021 26 / 42

Water Flow in a Channel Network

Fluvial and Torrential regime: the case of an incomingcanal

λ1 < 0

λ1 > 0

λ1 < 0

q = h√gh

admissible junction values

u0i

Maya Briani (IAC-CNR) Shallow water equations in channel networks May 5 - 2021 27 / 42

Water Flow in a Channel Network

Fluvial and Torrential regime: outgoing canal

u0j

λ2 > 0

λ2 > 0

λ2 < 0

Figure: |Fr| < 1.Maya Briani (IAC-CNR) Shallow water equations in channel networks May 5 - 2021 28 / 42

Water Flow in a Channel Network

The case study of a simple network

We consider a fictitious network formed by two canals intersecting at onesingle point, which artificially represents the junction.

u01 u02

J

(u∗1, u∗2)

I Conservation of Mass q∗1 = q∗2I Junction Riemann Problem: u∗1 ∈ N (u01), u

∗2 ∈ P(u02)

I we need ? additional conditions

Maya Briani (IAC-CNR) Shallow water equations in channel networks May 5 - 2021 29 / 42

Water Flow in a Channel Network

The case study of a simple network: 1 → 1 Junction

I Conservation of Mass q∗1 = q∗2I Junction Riemann Problem: u∗1 ∈ N (u01), u

∗2 ∈ P(u02)

I we need additional conditions ...

In this simple junction, the natural assumption (consistent with thedynamic of shallow-water equations on a single canal) should be to assumethe conservation of the momentum:

(q∗1)2

h∗1+

1

2g(h∗1)

2 =(q∗2)

2

h∗2+

1

2g(h∗1)

2.

Maya Briani (IAC-CNR) Shallow water equations in channel networks May 5 - 2021 30 / 42

Water Flow in a Channel Network

The case study of a simple network: 1 → 1 Junction

From the conservation of the momentum and q∗1 = q∗2(h2h1

)3

−(2F2

1 + 1)(h2

h1

)+ 2F2

1 = 0

and we have two possible relations for the heights values at the junction:

h∗1 = h∗2 (equal heigths) orh∗2h∗1

=1

2

(−1 +

√1 + 8F2

1

)Let us assume equal water heights, no jump at the junction

h∗1 = h∗2

Then u∗1 = u∗2 = u∗ and the solution is identified (if exists) by theintersection between the two admissible regions N (u01) and P(u02)!

Maya Briani (IAC-CNR) Shallow water equations in channel networks May 5 - 2021 31 / 42

Water Flow in a Channel Network

1 → 1 Junction: Fluvial → Fluvial

ur

ulu∗↙

NB(ul)

PB(ur)

Maya Briani (IAC-CNR) Shallow water equations in channel networks May 5 - 2021 32 / 42

Water Flow in a Channel Network

1 → 1 Junction: Torrential → Fluvial

ur

ul ul

↑u∗

NB(ul)

PA(ur)

Maya Briani (IAC-CNR) Shallow water equations in channel networks May 5 - 2021 33 / 42

Water Flow in a Channel Network

1 → 1 Junction: Torrential → Fluvial

urul

ul ∈ P(ur)

ul

N (ul)

P(ur)

Maya Briani (IAC-CNR) Shallow water equations in channel networks May 5 - 2021 34 / 42

Water Flow in a Channel Network

1 → 1 Junction: Torrential → Torrential

ur

ul

ul /∈ P(ur)ul

N (ul)

P(ur)

No solution!

Maya Briani (IAC-CNR) Shallow water equations in channel networks May 5 - 2021 35 / 42

Water Flow in a Channel Network

The case study of a simple network: 1 → 1 Junction

Assuming h∗1 = h∗2 the solution does not always exist ...

I Assuming h∗1 = h∗2 the solution does not always exist ...

I For h∗1 6= h∗2 we get new possible solutions at the junction ... thecases Torrential→ Fluvial and case Torrential→ Torrential may admitsolution even if their admissible regions have empty intersection in thesubcritical region.

Consistency with the case of a single canal:for appropriate values of (hl, ql), for Torrential→ Fluvial we get the samesolution considering our simple network as a simple canal, i.e. we get astationary shock at the virtual junction called hydraulic jump characterizedindeed by the conservation of the momentum in the transition from asupercritical to subcritical flow.

Maya Briani (IAC-CNR) Shallow water equations in channel networks May 5 - 2021 36 / 42

Water Flow in a Channel Network

Fr > 1 Fr < 1→

Fr > 1 Fr < 1

Figure: Numerical test case for the configuration given in Fig. ??.

Maya Briani (IAC-CNR) Shallow water equations in channel networks May 5 - 2021 37 / 42

Water Flow in a Channel Network

The extension to more complex network is still an Open Problem!

Maya Briani (IAC-CNR) Shallow water equations in channel networks May 5 - 2021 38 / 42

Water Flow in a Channel Network

Conclusion

I Two regimes exist for this hyperbolic system of balance laws: thefluvial, corresponding to eigenvalues with different sign, and thetorrential, corresponding to both positive eigenvalues

I After analyzing the Lax curves for incoming and outgoing canals, weprovide admissibility conditions for Riemann solvers, describingpossible solutions for constant initial data on each canal.

I The simple case of one incoming and outgoing canal is treatedshowing that, already in this simple example, regimes transitionsappear naturally at junctions.

Maya Briani (IAC-CNR) Shallow water equations in channel networks May 5 - 2021 39 / 42

Water Flow in a Channel Network

Further work

I Open canals flow with fluvial to torrential phase transitions on morecomplex networks

I Condition at the node depending on the geometry of the network andcomparison with 2D simulations (joint work with M. Ribot and G.Puppo)

I Problems with non constant bottom level and non constant widthchannels on networks

Maya Briani (IAC-CNR) Shallow water equations in channel networks May 5 - 2021 40 / 42

Water Flow in a Channel Network

Bibliography

M. Briani, G. Puppo, M. Ribot, Angle dependence in couplingconditions for shallow water equations at canal junctions. Preprint2021 hal-03196295. Submitted.

Briani, M.; Piccoli, B. Open canals flow with fluvial to torrential phasetransitions on networks. Networks & Heterogeneous Media, 2018, 13(4) : 663-690.

Briani, M.; Piccoli, B.; Qui, J. M. Notes on RKDG methods forshallow-water equations in canal networks. Journal of ScientificComputing 68(3) (2016).

Maya Briani (IAC-CNR) Shallow water equations in channel networks May 5 - 2021 41 / 42

Thank you very much for your attention

Maya Briani (IAC-CNR) Shallow water equations in channel networks May 5 - 2021 42 / 42